24 replies to this topic

[sim_sim]

Deal all,

 

I am trying to calibrate an atmospheric distillation column using the follwing method:

 

1. I start to converge the tower, considering 100% efficiency for all trays

2. I choose some performance specifications such as:

   .. The circulating rate and delta temperature of each pumparound , with exception to one (diesel pumpaournd)

    . The d86 5% and D86 95% of OVHD product;

    . D86 95%  of side products;

   . Reflux ratio

 

. With these specs, the flow rates of the OVHD, side products and bottom product are calculated by Petro-Sim.

 

3. After converging the tower, with the new specs, I start to change the number of trays in each section, starting from bottom to the top, and changing the number of trays in the side strippers too.

4. When the relative errors of some variables (products flow rates, temperatures in some trays, D86 5% of side products) are less than 10 %, I consider that the column is calibrated and then I start  the validation.

 

Well, I am facing some problems here, because when I start the validation  the results of the products  flow rates are quite different than the operation. I assume that the calibration is not properly done…

 

Can you give me your opinion about the specs, and the procedure that  I am using? Should I have to choose different specs (perhaps the top or bottom flow rate instead of another one)?

 

 

Best regards,

 

Bruno 

Edited by Bruno_Sim, 04 November 2013 - 05:32 AM.

[PingPong]

Are you trying to reproduce the operation of an existing column by simulation in Petro-Sim?

 

In that case: do not specify ASTM D86 temperatures, but specify actual product quantities.

ASTM D86 temperatures predicted by simulators are very inaccurate.

 

And why do you not also specify the diesel PA by rate and delta temperature?

 

But the most important input to your simulation is the TBP of the crude oil feed. If the hydrocarbon input is not correct then you can forget ever getting the correct hydrocarbon output. Read also this topic: http://www.cheresour...cdu-simulation/

[sim_sim]

Thank you for your quick reply.

 

Yes, I am trying to reproduce the operation of an existing column by simulation. 

I am using ASTM D86 temperatures because I am trying to build a model to predict the yields of the products, and so I was thinking in leaving the side products, top and bottom flow rates as degrees of freedom, and in that way the flow rates are calculated by the simulator.

 

I was not aware that the Petro-Sim does not calculate the ASTM D86 temperatures with accuracy.

 

Also, I don't specify the diesel PA rate because  I don't have a precise flow meter for that (the current has some problems) but I specify the delta temperature of diesel PA.

 

What specs do you suggest for, in my case, predict the yields of the different products?

 

Thank you,

 

Bruno

[PingPong]

I was not aware that the Petro-Sim does not calculate the ASTM D86 temperatures with accuracy.

I am no familiar with Petro-Sim, but no such correlation used by any simulator is accurate. API has changed their correlations several times in the past decades, so that says enough....

 

 

 

What specs do you suggest for, in my case, predict the yields of the different products?

You should first read that other topic about CDU modelling.

You can predict future product yields simply based on today's column TBP cutpoints. And today's TBP cutpoints simply follow from today's product quantities versus today's crude assay. You don't really need a simulator for that, you can do that by hand.

A simulator will however also predict vapor/liquid trafic in the column, and therefor tray loadings (% flood), and duties of feed heater, PA exchangers and overhead condensors. That's the real point of doing a simulation, but if you are not interested in duties, but only in product yields of a certain crude: simply apply today's column TBP cutpoints to any crude assay.

Edited by PingPong, 04 November 2013 - 10:16 AM.

[sim_sim]

Hi,

 

I just read the topic you suggested and now I understand that my choices were not reasonable. As my understanding I should not choose the ASTM D86 % cut points as specifications and instead I should choose the product flow rates. Should I choose top and bottom flow rates too?

 

I chose the specifications that I mentioned based on the book Refinery Process Modelling (http://www.amazon.co...s/dp/0967927471) and it refers that the one of the specifications should be D85 95% distilled of all product liquids except atmospheric gas oil...

 

Can you be kind to suggest some supporting literature to simulation of CDUs?

 

Concerning your last comment, the reason why I am trying to calibrate this column is for in the future to be able to do hidraulic tests (like %flooding) and be able to optimise the column and improve the column efficiency (in terms of energy mainly).

 

 

Thank your for all your help,

 

Bruno

[PingPong]

To be useful your model must be able to reproduce today's operation. That means that it must reproduce the heat and mass balance of the present operation. Therefor first specify the crude feed rate and the product rates of all products, except one (that follows automatically from the overall mass balance). And make all PA duties as in present operation. Rhen adjust the number of trays in each column section such that you approach the D86 temperatures from present operation. I don't know what methods in Petro-Sim are available to calculate the D86 temperatures, but try them all. However keep the tray efficiencies realistic, so choose the method that closest reproduces present D86 at realistic tray efficiencies, or fix tray efficiencies and choose that D86 method that closest approaches today's D86 during present operation. The D86 deviations are then deltas which you should take into account when you later use the model for a different operation than at present. For example: if you know that for gasoil your model calculates 7 ^oC too high for its 95% D86, and you want a 95% spec of 360 ^oC, then you spec 367 ^oC in the model (or you adjust the gasoil sidedraw rate manually until you get about 367 ^oC at 95% D86). Same for other products.

 

 

 

... the book Refinery Process Modelling ... and it refers that the one of the specifications should be D85 95% distilled of all product liquids except atmospheric gas oil...

Except gasoil? That's a strange advice, as that normally is a very important spec for gasoil.

Could you scan and upload the page which states that?

Edited by PingPong, 04 November 2013 - 04:44 PM.

[sim_sim]

Hi,

 

Thank you for your reply. I will make a calibration of the tower today, this time taking in account your advices.

.

 

 

Rhen adjust the number of trays in each column section such that you approach the D86 temperatures from present operation. I don't know what methods in Petro-Sim are available to calculate the D86 temperatures, but try them all. 

Petro-Sim uses API1974, API1987, API 1994 and Edmister Okamoto 

 

 

Except gasoil? That's a strange advice, as that normally is a very important spec for gasoil.

Could you scan and upload the page which states that?

Yes, please find the attached file.

 

 

Thanks,

 

Bruno

Attached Files

•  performance specifitations.pdf   119.9KB   94 downloads

[PingPong]

Okay, now I see where in the book it is mentioned. Overlooked it yesterday evening.

 

Note however that a few lines lower the Gasoil 95% distilled is mentioned as one of the possible specs. As overflash is usually unknown in an existing crude column, and simulated draw temperature can be very different from actual (see also Kaes' remarks on draw temperatures of PA's), the D86 95% is really the only spec to be used, unless one specifies the gasoil draw rate based on TBP cutpoints of actual operation and adjusts heater outlet temperature to get acceptable Gasoil D86 95% (by increasing overflash). With heater outlet temperature it is the same prioblem as with column draw temperatures: simulated temps can differ quite from actual temps. Depends on the thermo set used, and also on your choice of methods to determine Tc, Pc, omega and MW of the crude pseudofractions. It can take a lot of trials and time before you are able to accurately reproduce actual operation in your simulation model.

[sim_sim]

Hi,

 

 Thank you for the help you gave. I am testing the method we discussed. I am having some success, but your are right when you say it can take a lot of trials to achieve a satisfactory result.

Can you help with one more thing?

I am also trying to do a simulation (with the same column and same configuration) that estimates the stripping steam that the process needs to achieve the desired products, but I found that the simulation didn't even run or either start calculate...I am doing the exact thing I did till now (choosing the same specifications discussed above) but I left in blank the flow of stripping steam...

Is there a away to the simulator to estimate the stripping steam of the column?

 

Thank you

 

Bruno

[PingPong]

Stripping steam in sidestrippers if to assure that flaspoint of stripper bottoms is according to specification.

 

Stripping steam in column bottoms is to assure that flashpoint of atmospheric residue is according to specification, but also to recover gasoil from the residue.

 

The simulator is not going to use the "right" amount of stripping steam by itself. You have to specify something, for example stripper bottoms flashpoint. Or better: do runs with different amounts of stripping steam, and record flashpoints, ASTM D86 data, and gasoil in residue. In that way you get a good insight in the sensitivity of the product properties in relation to the strip steam quantity.

[sim_sim]

Thank you once again.

 

I will consider your advices.

 

Best Regards,

 

Bruno

[sim_sim]

Hi!

 

I am trying to calibrate the column for some operation days considering the advices you mentioned but I am noticing that is very difficult to match the D86 cuts, mainly the 5% cut of Kero and Diesel. Also the value of the temperature of the top of the column is most of the times less than the real values. The relative errors that I am having for these specs are more than 10%.

The specs I am using to perform the simulation are the flow rates of the products, except the top flow rate, the PAs flow rates except Diesel PA (because the flow meter here is not consistent), the reflux flow rate and the 5%D86 of the distillate.

Do you know what can I do to try to match the top temperature for the column? It seems to me that is mismatch of temperature is the origin of the other problems (D86 cut temperatures errors).

 

Thank you 

 

Bruno

[PingPong]

The 5 % D86 temperatures of Kero and Diesel depend op the effectiveness of the stripping.

I assume you specify the actual kg/h stripping steam as in present operation, so you need to increase number of theoretical stages in simulation in the sidestripper(s), and maybe also in the column between top & Kero and between Kero & Diesel drawoffs.

Note also that the calculation of the 5% D86 in any simulator is not very accurate. There are many methods which all give different results, and probably none is correct.

 

The colum top temperature is the dewpoint of the distillate (naphtha & lighter), plus reflux, plus overhead steam. If your distillate quantity, or reflux quantity or overhead steam quantity in simulation are not same as in actual operation then you obviously get wrong dewpoint and thus wrong top temperature. Look at total water condensed in overhead drum(s).

 

Flowrates measured in plant are often very inacurate, and moreover usually in volume, not mass. First tweak the operating flow rates such that you get a closing mass balance over the column, including any water in the crude (carryover from desalter). Without a closing operating mass balance, there is no point in trying to get a decent simulated mass balance.

 

As you do not specify top flow rate, any inaccuracy in other flow rates goes at expense of accuracy of resulting top flow rate. Maybe better to specify a top flow (to get "right" top temp) and leave residue flow open.

Edited by PingPong, 13 November 2013 - 03:30 PM.

[sim_sim]

Another problem I have, it's the flow rate of off-gas that I am having in the top condenser...The results of the simulation for the flow rate is more than 7ton/h...and it is to high (I don't have operation data for this one but for my experience I should not have that kind of flow rates in the field)...do you have any idea for what might be causing this problem? I tried your last advices and the relative errors for the cut were lower this time, but the flow rate of off-gas continues too high

 

Thank you

 

Bruno

Edited by Bruno_Sim, 14 November 2013 - 10:01 AM.

[PingPong]

Another problem I have, it's the flow rate of off-gas that I am having in the top condenser...The results of the simulation for the flow rate is more than 7ton/h...and it is to high (I don't have operation data for this one but for my experience I should not have that kind of flow rates in the field)

What is the Temperature and Pressure at the outlet of that condensor?

 

What sort of crude oil are you processing?

 

What is the crude oil flowrate?

[sim_sim]

What is the Temperature and Pressure at the outlet of that condensor?

Should be 60ºC and P=0.38 barg

 

What sort of crude oil are you processing?API gravity (dry)=  32.5 Specific Gravity (dry)=0.8628

 

What is the crude oil flowrate?464 ton/h

[PingPong]

Which crude is that? Arab Light? Or REBCO? Or what?

[sim_sim]

Actually it is a mix of angolan crudes

[PingPong]

I don't think that I have detailed assay data on Angolan crudes, but without the crude names I would not be able to find those anyway.

 

If it were an average middle east crude it can very well produce 7 t/h vapor at 60 ^oC / 0.38 barg from 464 t/h crude.

 

The light products that you get out of the condensor are the light fractions present in the crude feed that you use in the simulation.

In other words: you get out of the condensor what you put in as part of the crude feed to the simulation yourself.

 

Presumably you base the crude feed on Angolan crude assays that you have, so verify that those assays are reliable and are (still) representative of todays crude composition, and check that you entered the assay data correct in your simulation.

[sim_sim]

The crude names are: 

Kuito , Plutonio, Girassol and Antan (this is Nigerian)

 

I will try to check the real flow rate of off-gas and check if my simulation results are consistent with the operation data.

 

Thank you,

 

Bruno

[PingPong]

I am curious how your company makes a crude oil blend of 32.5 ^oAPI out of Kuito 19 ^oAPI , Plutonio 33 ^oAPI , Girassol 30 ^oAPI and Antan 26^oAPI , unless it is mainly Plutonio. Note that the actual crude blend composition has an impact on the vapor from the top condensor as the lights content of the crudes is very different.

 

I have no assay data on Antan, but the C₄ & Lighter content of the others is as follows:

 

Kuito . . . . . . 0.1 wt %

Plutonio .. . . 1.0 wt %

Girassol .. . . 0.7 wt%

Antan .. . . . . ???

 

In comparison: the Kuwait assay of that other CDU topic has 1.4 wt C₄ & Lighter, and that would result in 8 t/h vapor at 60 ^oC / 0.39 barg for 464 t/h crude.

 

Apparently Angolan crudes contain less lights than middle east crudes, so you should get less than 8 t/h vapor, but it will depend very much on the actual composition of the crude blend, in view of the big difference in Lights content.

 

But once again: what you get out of the simulation is what you put in the crude feed yourself.

And that applies not only to the vapour product, but also to the naphtha, kero and gasoil products.

[sim_sim]

Sorry my mistake...the API gravity is about 30.94º.

 

I have some information about Antan. the C₄ & Lighter content is:

 

Antan....1.1 wt%

 

and Girassol ....1.0wt% 

 

My offgas stream in the condenser has also C5 and C6...more than 10% m/m (according to simulation)...is it "normal" to have C5 and C6 in such quantitties?

 

Thank you

[PingPong]

The overhead drum (downstream condensor) is just a single stage flash, so do not expect a sharp separation between C4minus and C5plus. There will be a considerable amount of C4minus in the liquid, and there will be a considerable amount of C5plus in the vapor.

 

I only mentioned the C4minus content as an indicator of lights in crude, just to compare different crudes: the higher the C4minus content, the higher the vapor amount after the condensor.

 

What is the input in your simulation?

- Do you feed one crude blend? If so: what do you use as C1, C2, C3 and C4 content for that blend?

- Or do you feed the 4 crudes and mix them? If so: what is the blend ratio that you use?

[sim_sim]

What is the input in your simulation?

- Do you feed one crude blend? If so: what do you use as C1, C2, C3 and C4 content for that blend?

- Or do you feed the 4 crudes and mix them? If so: what is the blend ratio that you use?

 

Hi,

I use as feed a mix of 4 crudes, and the proportions are:

0.14 % KUI

5.38 % ANT

57.55 % GIR

36.93 % PUT

 

(%m/m)

 

 

Thanks

[PingPong]

Amount of vapor product from the drum downstream the condensor will very much depend on the amount of unstabilized naphtha product (excl. reflux) from that drum. In otrher words: it depends very much on the TBP cutpoint between this naphtha and the kero product.

 

At a TBP cutpoint of 150 ^oC , and consequently about 55 t/h unstab naphtha product, the vapour product could very well be 5 t/h.

At a TBP cutpoint of 180 ^oC , and consequently about 75 t/h unstab naphtha product, the vapour will be about 3 t/h.

That is assuming no cracked gas make in the crude heater.

 

However: If you have in your simulation used the cracked gas make that Kaes specifies in his book, then the amount of vapour from the condensor will be much higher. I would not use that cracked gas make. Moreover: note that his conversion between SCF/Bbl and Sm³/m³ is a factor 2 wrong!